System and method for condition monitoring of redox flow batteries using data analytics

• US 10,396,375 B2
• Filed: 03/24/2016
• Issued: 08/27/2019
• Est. Priority Date: 03/24/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Choudhury, Anamitra Roy, Dechu, Sampath, Kumar, Pratyush
• Primary Examiner(s): Nikmanesh, Seahvosh
• Application Number:
  US15/080,127
• Publication Number:
  US 20170279140A1
• Time in Patent Office: 1,251 Days
• CPC Class Codes: H01M 2250/10 Fuel cells in stationary systems, e.g. emergency power source in plant
• CPC Class Codes: H01M 8/008 Disposal or recycling of fuel cells
• CPC Class Codes: H01M 8/04305 Modeling, demonstration models of fuel cells, e.g. for training purposes
• CPC Class Codes: H01M 8/0432 Temperature; Ambient temperature
• CPC Class Codes: H01M 8/04477 of the electrolyte
• CPC Class Codes: H01M 8/04544 Voltage
• CPC Class Codes: H01M 8/04574 Current
• CPC Class Codes: H01M 8/0482 of the electrolyte
• CPC Class Codes: H01M 8/04992 characterised by the implementation of mathematical or computational algorithms, e.g. feedback control loops, fuzzy logic, neural networks or artificial intelligence
• CPC Class Codes: H01M 8/18 Regenerative fuel cells, e.g. redox flow batteries or secondary fuel cells
• CPC Class Codes: H01M 8/188 by recharging of redox couples containing fluids; Redox flow type batteries
• CPC Class Codes: H01M 8/20 Indirect fuel cells, e.g. fuel cells with redox couple being irreversible H01M8/18 takes precedence
• CPC Class Codes: H01M 8/225 Fuel cells in which the fuel is based on materials comprising particulate active material in the form of a suspension, a dispersion, a fluidised bed or a paste
• CPC Class Codes: Y02E 60/50 Fuel cells
• CPC Class Codes: Y02W 30/84 Recycling of batteries or fuel cells

Abstract

One embodiment provides a method for predicting maintenance of a redox flow battery, the method including: receiving, from a plurality of sensors, data regarding characteristics of the redox flow battery; weighting, using a processor, each of the characteristics to form an estimated state parameter for the redox flow battery; and determining, using the processor, a maintenance action for the redox flow battery using the estimated state parameter. Other aspects are described and claimed.

Claims

1. A method for predicting maintenance of a redox flow battery, the method comprising:

receiving, from a plurality of sensors, data regarding characteristics of the redox flow battery, wherein the characteristics identify the state of a parameter of the redox flow battery and comprise electrical characteristics, environmental characteristics, and battery level characteristics;

forming an estimated state parameter for the redox flow battery by weighting each of the characteristics and aggregating the weighted characteristics, wherein the estimated state parameter indicates an operational state of the redox flow battery;

the forming an estimated state parameter comprising utilizing a model that is trained using a plurality of sensor measurements from a plurality of different sensors measuring different characteristics of the redox flow battery and taken at known state parameters of the redox flow battery to weight characteristics and estimate state parameters, wherein to weight the characteristics is based upon an accuracy of a value provided by the sensor corresponding to the characteristic, the accuracy being determined using a machine learning algorithm;

determining, using the processor, a maintenance action for the redox flow battery using the estimated state parameter, wherein the determining comprises analyzing the estimated state parameter against historical work order data and in view of utilities operation information to identify a maintenance action corresponding to the estimated state parameter; and

providing a recommendation of the determined maintenance action.

2. The method of claim 1, wherein the estimated state parameter comprises state of charge of the redox flow battery.

3. The method of claim 2, comprising receiving operational history data for the redox flow battery;

wherein the determining takes into account the operation history data.

4. The method of claim 2, comprising receiving predicted operational load data for the redox flow battery;

wherein the determining takes into account the predicted operational load data.

5. The method of claim 2, wherein the data regarding characteristics of the redox flow battery are selected from the group consisting of:

temperature data, current data, voltage data, conductivity data, potentiometeric data, and optical data.

6. The method of claim 5, comprising:

training a regression model for estimating the state of charge using a conductivity characteristic of the redox flow battery;

training the regression model for estimating the state of charge using a potentiometeric characteristic of the redox flow battery;

training a threshold function that estimates an electrolyte color change with respect to the state of charge of the redox flow battery; and

utilizing temperature buckets to identify which of the plurality of sensors provides a most accurate estimate of the state of charge of the redox flow battery for a given temperature.

7. The method of claim 6, wherein the weighting comprises:

measuring a current temperature of the redox flow battery using a temperature sensor;

identifying a sensor of the plurality of sensors that provides the most accurate estimate of the state of charge at the current temperature;

determining if the estimate of the state of charge is within a predetermined range associated with electrolyte color change; and

responsive to determining if the estimate of the state of charge is within the predetermined range, performing a confirming measurement with an optical sensor.

8. The method of claim 1, wherein the maintenance action is selected from the group consisting of electrolyte remixing, electrode replacement, electrode cleaning, adding a reductant, adding inert gas, electrolyte refilling, tank replacement, pump replacement, leak fixing, and electrolyte modification.

9. The method of claim 1, wherein the redox flow battery is a vanadium redox flow battery.

10. An apparatus for predicting maintenance of a redox flow battery, the apparatus comprising:

at least one processor; and

a non-transitory computer readable storage medium having computer readable program code embodied therewith and executable by the at least one processor, the computer readable program code comprising;

computer readable program code that receives, from a plurality of sensors, data regarding characteristics of the redox flow battery, wherein the characteristics identify the state of a parameter of the redox flow battery and comprise electrical characteristics, environmental characteristics, and battery level characteristics;

computer readable program code that forms an estimated state parameter for the redox flow battery by weighting each of the characteristics and aggregating the weighted characteristics, wherein the estimated state parameter indicates an operational state of the redox flow battery;

the forming an estimated state parameter comprising utilizing a model that is trained using a plurality of sensor measurements from a plurality of different sensors measuring different characteristics of the redox flow battery and taken at known state parameters of the redox flow battery to weight characteristics and estimate state parameters, wherein to weight the characteristics is based upon an accuracy of a value provided by the sensor corresponding to the characteristic, the accuracy being determined using a machine learning algorithm;

computer readable program code that determines a maintenance action for the redox flow battery using the estimated state parameter, wherein the determining comprises analyzing the estimated state parameter against historical work order data and in view of utilities operation information to identify a maintenance action corresponding to the estimated state parameter; and

computer readable program code that provides a recommendation of the determined maintenance action.

11. A computer program product for predicting maintenance of a redox flow battery, the computer program product comprising:

a non-transitory computer readable storage medium having computer readable program code embodied therewith, the computer readable program code being executable by a processor and comprising;

computer readable program code that receives, from a plurality of sensors, data regarding characteristics of the redox flow battery, wherein the characteristics identify the state of a parameter of the redox flow battery and comprise electrical characteristics, environmental characteristics, and battery level characteristics;

computer readable program code that forms an estimated state parameter for the redox flow battery by weighting each of the characteristics and aggregating the weighted characteristics, wherein the estimated state parameter indicates an operational state of the redox flow battery;

the forming an estimated state parameter comprising utilizing a model that is trained using a plurality of sensor measurements from a plurality of different sensors measuring different characteristics of the redox flow battery and taken at known state parameters of the redox flow battery to weight characteristics and estimate state parameters, wherein to weight the characteristics is based upon an accuracy of a value provided by the sensor corresponding to the characteristic, the accuracy being determined using a machine learning algorithm;

computer readable program code that determines a maintenance action for the redox flow battery using the estimated state parameter, wherein the determining comprises analyzing the estimated state parameter against historical work order data and in view of utilities operation information to identify a maintenance action corresponding to the estimated state parameter; and

computer readable program code that provides a recommendation of the determined maintenance action.

12. The computer program product of claim 11, wherein the estimated state parameter comprises state of charge of the redox flow battery.

13. The computer program product of claim 12, comprising computer readable program code that receives operational history data for the redox flow battery;

wherein the computer readable program code that determines a maintenance action takes into account the operation history data.

14. The computer program product of claim 12, comprising computer readable program code that receives predicted operational load data for the redox flow battery;

wherein the computer readable program code that determines a maintenance action takes into account the predicted operational load data.

15. The computer program product of claim 12, wherein the data regarding characteristics of the redox flow battery are selected from the group consisting of:

temperature data, current data, voltage data, conductivity data, potentiometeric data, and optical data.

16. The computer program product of claim 15, comprising:

computer readable program code that trains a regression model for estimating the state of charge using a conductivity characteristic of the redox flow battery;

computer readable program code that trains the regression model for estimating the state of charge using a potentiometeric characteristic of the redox flow battery;

computer readable program code that trains a threshold function that estimates an electrolyte color change with respect to the state of charge of the redox flow battery; and

computer readable program code that utilizes temperature buckets to identify which of the plurality of sensors provides a most accurate estimate of the state of charge of the redox flow battery for a given temperature.

17. The computer program product of claim 16, wherein the computer readable program code that weights comprises:

computer readable program code that measures a current temperature of the redox flow battery using a temperature sensor;

computer readable program code that identifies a sensor of the plurality of sensors that provides the most accurate estimate of the state of charge at the current temperature;

computer readable program code that determines if the estimate of the state of charge is within a predetermined range associated with electrolyte color change; and

computer readable program code that, responsive to determining the estimate of the state of charge is within the predetermined range, performs a confirming measurement with an optical sensor.

18. The computer program product of claim 11, wherein the maintenance action is selected from the group consisting of electrolyte remixing, electrode replacement, electrode cleaning, adding a reductant, adding inert gas, electrolyte refilling, tank replacement, pump replacement, leak fixing, and electrolyte modification.

19. The computer program product of claim 11, wherein the redox flow battery is a vanadium redox flow battery.

20. A system for predicting maintenance of a redox flow battery, the method comprising:

a redox flow battery operatively coupled to a commercial power grid; and

an apparatus comprising;

at least one processor; and

a non-transitory computer readable storage medium having computer readable program code embodied therewith and executable by the at least one processor, the computer readable program code comprising;

computer readable program code that receives, from a plurality of sensors, data regarding characteristics of the redox flow battery, wherein the characteristics identify the state of a parameter of the redox flow battery and comprise electrical characteristics, environmental characteristics, and battery level characteristics;

computer readable program code that forms an estimated state parameter for the redox flow battery by weighting each of the characteristics and aggregating the weighted characteristics, wherein the estimated state parameter indicates an operational state of the redox flow battery;

the forming an estimated state parameter comprising utilizing a model that is trained using a plurality of sensor measurements from a plurality of different sensors measuring different characteristics of the redox flow battery and taken at known state parameters of the redox flow battery to weight characteristics and estimate state parameters, wherein to weight the characteristics is based upon an accuracy of a value provided by the sensor corresponding to the characteristic, the accuracy being determined using a machine learning algorithm;

computer readable program code that determines a maintenance action for the redox flow battery using the estimated state parameter, wherein the determining comprises analyzing the estimated state parameter against historical work order data and in view of utilities operation information to identify a maintenance action corresponding to the estimated state parameter; and

computer readable program code that provides a recommendation of the determined maintenance action.